Per-Credit vs Subscription Pricing: My 8-Month Mistake

I ran AdLoft AI on a per-credit model for eight straight months. Users bought packs of 10 or 50 credits, each credit unlocking one ad creative. On paper it looked flexible. In practice it created friction that killed retention.

Per-credit approach

Pros were obvious at first. New users could test without committing to a monthly bill. I could point to low entry cost during sales calls. Tracking was simple: one creative, one credit deducted.

Cons showed up fast. Power users burned through credits in a single campaign and then paused while they debated buying more. Casual users treated the balance like a scarce resource and generated fewer variations than they needed. Churn spiked every time a balance hit zero. Support tickets asked the same question weekly: "how many credits do I need for my next batch?"

Subscription approach

After eight months I flipped to flat monthly plans. Three tiers, unlimited generations within fair-use limits, and a hard cap only on the lowest plan.

Pros appeared immediately. Users stopped worrying about running out mid-campaign. Average creatives produced per account rose 3x. Monthly recurring revenue became predictable instead of lumpy pack purchases. The mental model shifted from "pay to try" to "pay to run my ads."

Cons exist. Some very light users now feel the base price is high for the few creatives they need. I lost a slice of price-sensitive testers who only wanted one or two images. Refunds increased slightly in the first month while people adjusted.

What the data showed

The switch was not a clean win. Revenue per active account rose, but total sign-ups dropped 18 % because the low-friction trial credit pack disappeared. Net revenue still grew because retained users stayed longer and upgraded. The biggest surprise was support load: ticket volume fell by half once the credit meter vanished.

A real example came from a small fashion store testing seasonal ads. Instead of counting credits they now run dozens of variations in one afternoon. They found the right angle faster and their ROAS improved. That kind of workflow only happens when the cost of another generation feels close to zero.

I still keep a one-time top-up option for users who need a short burst outside their plan. It satisfies the edge case without returning to the old meter anxiety.

The lesson is simple. Per-credit pricing optimized for acquisition but punished usage. Subscription pricing optimized for usage and let revenue follow. Neither model is universally better; the right choice depends on how often your customers actually need the output. I just happened to pick the wrong default for the first eight months.

one-click ad creative tool helped surface this pattern because fashion sellers generate high volumes quickly once friction drops.

AdLoft AI is an AI-powered ad creative generator that turns product photos into professional ad creatives instantly — no designer, no prompt engineering.